[Xen-devel] 答复: 答复: 答复: 答复: 答复: Help: a xen crash of 4.8.2 version/////答复: Is there a faster way to restore Virtual machine status in Xen?

["Chenjia (C)" <chenjia09@xxxxxxxxxx>]

Dear xen expert：
        From the blocked information, we found that the CPU 14 and CPU 19 is 
blocked, and the call trace is mainly about:
[68915.792526]  [<ffffffff811024f8>] on_each_cpu+0x28/0x60
[68915.792540]  [<ffffffff813ed7e1>] decrease_reservation+0x261/0x2f0
[68915.792555]  [<ffffffff813edfc6>] alloc_xenballooned_pages+0xe6/0x180
[68915.792571]  [<ffffffff813ec211>] gnttab_alloc_pages+0x11/0x40
[68915.792586]  [<ffffffffa049c15c>] gntdev_ioctl+0x23c/0x750 [xen_gntdev]
[68915.792608]  [<ffffffff81213d43>] do_vfs_ioctl+0x2e3/0x4c0
[68915.792631]  [<ffffffff81213f94>] SyS_ioctl+0x74/0x80
[68915.792644]  [<ffffffff815f57c3>] entry_SYSCALL_64_fastpath+0x1e/0xca

[69175.786405] Call Trace:
[69175.786418]  [<ffffffff8118b1e8>] pagecache_get_page+0x168/0x1d0
[69175.786471]  [<ffffffffa02685b9>] prepare_pages+0xa9/0x1b0 [btrfs]
[69175.786508]  [<ffffffffa0269342>] __btrfs_buffered_write+0x212/0x680 [btrfs]
[69175.786543]  [<ffffffffa026cfe3>] btrfs_file_write_iter+0x183/0x5c0 [btrfs]
[69175.786560]  [<ffffffff81201546>] __vfs_write+0xb6/0x100
[69175.786574]  [<ffffffff81201b3d>] vfs_write+0x9d/0x190
[69175.786587]  [<ffffffff81202812>] SyS_write+0x42/0xa0
[69175.786599]  [<ffffffff815f57c3>] entry_SYSCALL_64_fastpath+0x1e/0xca
[69175.793881] DWARF2 unwinder stuck at entry_SYSCALL_64_fastpath+0x1e/0xca

        What the reason maybe to cause this problem? Is there some way to 
Prevent the kernel from doing similar processing and avoiding this problems? 
Looking forward to your reply,thank you!

Best Regards


-----邮件原件-----
发件人: Chenjia (C) 
发送时间: 2018年7月9日 10:12
收件人: 'Juergen Gross' <jgross@xxxxxxxx>
抄送: zhaobingjian <zhaobingjian@xxxxxxxxxx>; Shentao (Terry) 
<shentao8@xxxxxxxxxx>; Yaoshaomin <yaoshaomin@xxxxxxxxxx>; Zhuxiaolin (A) 
<zhuxiaolin09@xxxxxxxxxx>; wangxu (R) <wangxu0@xxxxxxxxxx>; xen-devel 
<xen-devel@xxxxxxxxxxxxxxxxxxxx>; yuanjinfeng <yuanjinfeng@xxxxxxxxxx>; Hutao 
(C) <hutao09@xxxxxxxxxx>; Jugang (David, Enterprise Network) 
<jugang09@xxxxxxxxxx>; Jan Beulich <JBeulich@xxxxxxxx>
主题: 答复: 答复: 答复: 答复: [Xen-devel] 答复: Help: a xen crash of 4.8.2 version/////答复: 
Is there a faster way to restore Virtual machine status in Xen?

Dear Juergen:
        We have disable DPDK and add conring_size=1M on the server(xen 4.8.3 
with dom0 linux kernel 4.4.121-92.85-default), and we still got the blocked 
information, please see the attach file to help us. Thank you!

Best Regards

-----邮件原件-----
发件人: Juergen Gross [mailto:jgross@xxxxxxxx] 
发送时间: 2018年7月6日 15:50
收件人: Chenjia (C) <chenjia09@xxxxxxxxxx>; Jan Beulich <JBeulich@xxxxxxxx>
抄送: zhaobingjian <zhaobingjian@xxxxxxxxxx>; Shentao (Terry) 
<shentao8@xxxxxxxxxx>; Yaoshaomin <yaoshaomin@xxxxxxxxxx>; Zhuxiaolin (A) 
<zhuxiaolin09@xxxxxxxxxx>; wangxu (R) <wangxu0@xxxxxxxxxx>; xen-devel 
<xen-devel@xxxxxxxxxxxxxxxxxxxx>; yuanjinfeng <yuanjinfeng@xxxxxxxxxx>; Hutao 
(C) <hutao09@xxxxxxxxxx>
主题: Re: 答复: 答复: 答复: [Xen-devel] 答复: Help: a xen crash of 4.8.2 version/////答复: 
Is there a faster way to restore Virtual machine status in Xen?

On 06/07/18 08:27, Chenjia (C) wrote:
> Dear Juergen:
>            We will follow your suggestion: unload DPDK, then test a again.
> 
>       Our server have 24 vcpu, and if we press '0' it only show 10 vcpu's 
> dump message,  is there a way to show more dump message?

Looking more at this I guess the console ring buffer is too small.

Please add

conring_size=1M

to your hypervisor boot parameters.


Juergen
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel